JPH01191199A - Voice input device - Google Patents

Abstract

PURPOSE:To easily and efficiently input word information of a name, etc., by recognizing a syllable which is brought to voice input, deriving and displaying a prescribed number of syllable candidates, collating the word information stored in a word dictionary and deriving a work candidate intended by a syllable train. CONSTITUTION:By using a syllable recognizing part 3, a syllable which is brought to utterance input by an operator is recognized, a prescribed number of syllable candidates are derived and displayed on a display part 5, and offered for a selecting instruction by the operator. Subsequently, a syllable train shown by the syllable which is brought to selecting instruction and word information stored in a word dictionary are collated successively from its head part and a word candidate intended by the syllable train is derived, and by displaying this word candidate, a confirmation for inputting the information intended by the operator is urged. In such a way, a word of a name, etc., being an input purpose is delimited to syllables and brought to utterance input, and by only selecting and designating successively a recognition candidate to its syllable, the word candidate intended by the syllable train whose input has been decided definitely is derived successively and displayed. Accordingly, a troublesome operation for the selection and designation, etc., of the Japanese syllabary table becomes unnecessary.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to a voice input device that can efficiently specify and input words input by an unspecified speaker.

(Prior Art) When operating an automatic transfer device, it is usually necessary to input the client's name, beneficiary's name, transfer financial institution name, text name, beneficiary's account number, transfer amount, etc. The account number, amount to be transferred, etc. can be directly input using the so-called numeric keypad, so the operation is easy.

However, for the name of a financial institution and its main body ζ〆name, it is limited to the extent that the pronunciation information must be entered as a kana string, etc., so for example, it is necessary to enter the initial letter of the financial institution name from the 50-syllabary table displayed on the terminal screen. In order to enter the name of the financial institution by selecting and inputting the corresponding characters, there is a problem in that it takes a considerable amount of time to input the characters from the 50-syllabary table, and the input operation is very troublesome.

(Problems to be Solved by the Invention) As described above, with the input devices built into conventional automatic transfer devices, etc., inputting a wide variety of financial institution names, etc. requires extremely troublesome operations. There was a problem that the time required for operation was extremely long.

The present invention was made in consideration of these circumstances, and its purpose is to easily and efficiently input word information such as a wide variety of names such as the name of a financial institution and the name of the body of the voice. The purpose of the present invention is to provide a voice input device that can perform the following tasks.

[Structure of the Invention] (Means for Solving the Problems) The present invention uses a syllable recognition unit to recognize a syllable inputted by an operator, obtains a predetermined number of syllable candidates, and displays them on a display unit. Provides selection instructions by the operator. Then, the syllable string indicated by the selected syllable and the word information stored in the word dictionary are sequentially compared from the beginning to find the word candidate intended by the syllable string, and this word candidate is displayed and the word information stored in the word dictionary is sequentially compared. This system is characterized in that the operator is prompted to confirm the input of the intended information.

(Operation) According to the present invention, by simply inputting a word such as a name to be input by dividing it into syllables, and sequentially selecting and specifying recognition candidates for the syllable, the input confirmed syllable string becomes the intended word. Candidates are searched for and displayed in sequence. Therefore 50
To easily and efficiently obtain and input a word to be input by simply inputting the word voice in sequence while dividing it into syllables, without any troublesome operations such as selecting and specifying a sound table.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram of a voice input device according to an embodiment of the present invention, which is incorporated into an automatic transfer device.

In this type of automatic transfer device, information on the destination and source of money to be transferred is inputted together with information on the name of the financial institution, and furthermore, the amount to be transferred and the like are inputted. This device, which is built into such an automatic transfer device, can input names (word information) such as the above-mentioned financial institution name with just a simple command operation.
Moreover, it is intended to enable easy input without the need for troublesome key input operations as seen in conventional devices.

Therefore, this device uses voice as an information input medium, and realizes efficient information input by simply selecting and instructing the recognition results of input voice.

The present device is mainly composed of a control section 1 made up of, for example, a CPU, and roughly includes a speech recognition section and a word matching section as described below.

The speech recognition section acoustically analyzes the monosyllables, which are divided into monosyllabic units and inputted sequentially, using the speech input section 2, and uses the characteristic patterns of the acoustically analyzed monosyllables and the Japanese registered in the monosyllable dictionary 3. The word monosyllable standard pattern is compared with the word monosyllable standard pattern in the speaker-independent monosyllable recognition unit 4 to obtain monosyllable candidates. This speaker-independent monosyllabic recognition calculates the degree of matching between the so-called standard Japanese 1-syllable pattern and the characteristic pattern of the input monosyllable, for example - the top five monosyllabic categories with the highest political change. is obtained as a monosyllable candidate.

The monosyllable candidates obtained in this manner are displayed on the display unit 5, and the operator is provided with an instruction to select the correct one.

Here, the display section 5 and the selection section 6 for inputting selection information for the information displayed on the display section 5 are, for example, a laminated liquid crystal display 5 as shown in FIG.
a and a transparent tablet 8a. As will be described later, a display screen is formed as shown in FIG. 4, and when a specific position on the display screen is specified, information specified by the specified position is input.

Specifically, as shown in FIG. 2, the screen information is developed on the display memory 11 and displayed on the liquid crystal display 5a according to the display screen information, and the specified position coordinates are indicated on the transparent tablet (Ba). The coordinates are determined by the coordinate detection section 12. Then, the instruction information determining section 13 compares the instruction coordinates with the information on the display screen expanded on the display memory 11 to determine the information inputted by the instruction.

Although an example in which a transparent tablet 6a is used as the selection section 6 is shown here, it is also possible to use a so-called touch data unit or the like.

Accordingly, the word matching section 7, which is a word matching section in FIG. Matching processing is performed between the input target word and the input target word.

This word matching process sequentially determines whether or not there is a word that includes the input confirmed monosyllable or monosyllable string in the initial part, and matches the input confirmed monosyllable or monosyllable string to the initial letter. When a word included in a character part is detected, the word is extracted as a candidate word to be input.

The word candidates obtained in this way are displayed on the display section 5, and the operator is prompted to confirm the input. Confirmation information is sent from the selection unit 6 to the displayed word candidates,
Alternatively, when selection information is input, the designated word candidate is determined as the word to be input, and this is inputted.

FIG. 3 shows an example of a word input processing procedure by the control section 1 in the apparatus configured as described above, and FIG. 4 shows an example of a display screen at that time.

An example of the operation of this device will be described according to the processing procedure shown in FIG. 3. When the device is started, first, a parameter i for monosyllable input is initialized (i-1) (step a). Then, a display screen as shown in FIG. 4(a) is formed, and the operator is prompted to input the sound of a word such as the name of a financial institution in monosyllable units.

After such initialization processing, a monosyllable uttered by the operator is input via the voice input unit 2 (step b), and the characteristic pattern of the input monosyllable and the 101 registered in the monosyllable recognition dictionary 3 are input. Monosyllable recognition processing is performed between each standard pattern of monosyllables (step C). Then, the degree of matching obtained by this monosyllable recognition process is compared with each other, and the top five monosyllables with the highest political change are determined as candidates, and these are displayed in a predetermined monosyllable candidate display area α on the display section 5. (step d). It should be noted that if no monosyllabic candidates are found up to the fifth rank, it goes without saying that only the found monosyllabic candidates are displayed.

In this state, the operator waits for an instruction input for the monosyllable candidate. Then, either one of the monosyllabic candidates displayed on the display section B is selected via the selection section B, or the correct answer is not included in the displayed monosyllabic candidates, so the monosyllable is re-inputted, that is, reworded. It is determined whether an instruction is given (step e). When a rephrasing is instructed, the process from step b described above is repeatedly executed, and the recognition process for the re-input monosyllable is performed in the same way.

On the other hand, if one of the displayed monosyllable candidates is specified, the monosyllable candidate is determined as the input monosyllable (step f), and it is placed at the monosyllable position specified by the value of the parameter i. Memorize temporarily. At the same time, this input confirmed monosyllable is displayed in a predetermined monosyllable display area β on the display section 5 according to the monosyllable input position (
Step g).

Thereafter, words are collated as described above for the monosyllable or monosyllable string input and confirmed as described above (step h). In other words, it is checked whether the input target word (for example, the name of a financial institution) that includes the confirmed input monosyllable or monosyllable string at the beginning is registered in the word dictionary 8, and the input target word that meets this condition is selected as a word candidate. Extract. This word candidate is then displayed in a predetermined word candidate display area γ on the display unit 5. It is determined whether or not confirmation information can be obtained for the word candidate displayed in a stepwise manner (step j). This determination is made by determining whether or not one of the word candidates is selected via the selection unit 6, or if only one word candidate is requested, whether or not a confirmation instruction is input. It is done as follows.

Based on this determination, word candidates for the input purpose are displayed,
When confirmation information (selection information) for this is selected and instructed from the selection unit 6, the word candidate is determined as the word to be input (step No.). Then, input processing is performed for the determined word (step m), and preparation is made for the next word input processing.

Note that if the input for the searched word candidate is not confirmed, the parameter i is incremented (
+1)L (step k), the next monosyllable input is made. The processing procedure from step b described above is then repeatedly executed to perform word matching on the monosyllable string up to the input monosyllable.

Through the above processing procedure, when inputting the word "Toukiyo", for example, the monosyllable candidates are obtained as "to", "so", etc. by inputting a monosyllable, and for example, the first monosyllable is " ” is confirmed. Then, by matching words according to this monosyllable, corresponding words such as "Tokyo", "Tokai", "Tochigi", "Toyama", etc., are found.
Since the number of applicable words is large, it is not determined as a word candidate.

Then, when the next monosyllable ``u'' is uttered and input, this monosyllable is confirmed in the same way, and as a result, it is determined that the input monosyllable string is ``tou''. Then, matching words are narrowed down by word matching on this monosyllable string, and word candidates are determined as "Tokyo", "Tokai", etc. These word candidates are then displayed and subjected to input word confirmation processing.

The above processing is continued while inputting monosyllables in sequence until the target word is obtained as a word candidate. As a result, the target word information can be easily and efficiently inputted by inputting the target word into monosyllabic segments and confirming (selecting) the displayed monosyllable candidates and word candidates. becomes possible.

In addition, if the accuracy of word input by monosyllabic utterance is poor, specify whether the input voice is a male or female voice,
The monosyllabic standard pattern used for monosyllabic recognition may be switched.

If the recognition accuracy for input monosyllables is poor even after doing this, for example, change the display screen to one showing a 50-syllabary table for inputting kana, as illustrated in FIG. 4(b), It is also possible to configure the device so that the above-mentioned voice input can be backed up.

As described above, according to the present device, the target word can be determined and input very easily by inputting a single syllable of the target word, and the input efficiency is extremely high. Moreover, since the input is a monosyllable, even a single syllable including a sulcus sound such as "gyo" may be input only once. By the way, in the case of conventional kana input from the 50-syllabary table, three inputs are required as "ki Jr" and "Jr yo". Therefore, by inputting a single syllable, it is possible to greatly simplify the input processing.

Furthermore, according to the above-mentioned monosyllabic input, even if the "u" in "Tokyo J" is uttered as "o", the correct one can be easily selected from the monosyllabic candidates, making the word candidate extraction process more effective. Effects such as being able to perform

Furthermore, since the voice input information can be determined easily by operating switches on the display screen, the processing operation can proceed very smoothly.

Note that the present invention is not limited to the embodiments described above. Although the voice input device incorporated in an automatic transfer device has been described here, it can be similarly applied as a voice input device in various other devices (systems). Furthermore, the configuration example of the display screen, the display area for monosyllable candidates, word candidates, etc. may be determined according to the specifications. In particular, for voice input such as the name of a financial institution, since the type is determined in advance, the word information indicated by the input voice can be determined using a dictionary effect. In addition, the present invention can be implemented with various modifications without departing from the gist thereof.

[Effects of the Invention] As explained above, according to the present invention, it is only necessary to sequentially input the sounds of the target word as monosyllables.
The input operation can be greatly simplified, and it is possible to improve the input processing efficiency of words, which is largely dependent on the input efficiency of the operator, such as inputting information such as the transfer destination in an automatic transfer device. Many practical effects can be achieved.

[Brief explanation of the drawing]

The figure shows a voice input device according to an embodiment of the present invention that is incorporated into an automatic transfer device. FIG. 1 is a schematic configuration diagram of the embodiment device, and FIG. 2 is a configuration example of a display section and a selection section. FIG. 3 is a diagram showing an example of an input processing procedure, and FIG. 4 is a diagram showing an example of a display screen. DESCRIPTION OF SYMBOLS 1... Control unit, 2... Speech input unit, 3... Monosyllable recognition dictionary, 4... Unspecified speaker recognition unit, 5... Display unit, 6... Selection unit, 7・・・Word matching department, death...
Word dictionary. Applicant's agent Patent attorney Takehiko Suzue Figure 1 Figure 2

Claims (1)

[Claims] a syllable recognition section that recognizes a syllable input as a voice and obtains a predetermined number of syllable candidates; a display section that displays the predetermined number of syllable candidates obtained by the syllable recognition section; and a predetermined number of syllable candidates displayed on the display section. a selection unit for selecting and instructing one of the syllable candidates; a word dictionary storing word information to be input;
a word matching section that searches for a word candidate intended by the syllable string by matching the syllable string indicated by the selected syllable with word information stored in the word dictionary; and a word candidate found by the word matching section. 1. A voice input device comprising: means for displaying.